by Doug Coulter » Sun May 29, 2011 5:26 pm
When it's doing that -- most of the pulses look just like they do for the real thing, though some are "louder" and get clipped. I did an input voltage sweep while it was acting nuts and it starts right at the corona onset voltage (6-700 v), and just gets bigger and faster with more HV applied up to the limit on these of 2500v or so. Then it gradually cleans itself up (so far). If you start out at 2500, and turn the volts down, it quits for awhile, but them comes back. It's just like HV cleanup/conditioning in how it seems to act. Like dirt is flying around in there under electrostatic forces and making counts. It's real random, with bursts of faster and slower for awhile.
In fact, I just went down and tried it again. It counted like crazy for 2-3 minutes, then settled down to what I assume is the cosmic background rate -- every few seconds. The "fake" counts tend to be larger and flat topped, and wider. Then it seems to just go away. I suppose what I should do it hook it to an audio amp and monitor it over some hours to make sure this is merely a turn-on phenomenon. It'd really mess you up if you were integrating total counts and it happened during a fusor run - the fakes are faster than from all but a very hot fusor with the thing in close. This time the thing hadn't been touched since the last run other than the power switch, so it's not due to being shaken up.
When we first saw elevated counts after a run, I made some mods. I painted the inside of the coffee can, used a real tube type insulated plate cap for the tube, added a quartz tube insulator over the 50 meg R chain, and added a HV bypass cap (mainly used as a strain relief for the HV wires) coming in. Further I made a 400 pf coupling cap out of two 200 pf ones as stated in the paper you found. None of that made any difference at all - but did eliminate the theory that the extra counts were corona noise off the sharp pointed connector I'd used on the tube end to begin with, and now I know "it's as right as my hands can make it". None of these mods had any particularly noticeable effect -- other than I know I did it right now. The shielding we did makes this the most impervious to EMI detector in the house, BTW. Very nicely silent when it should be (and the tube is right). Jon made those nice flanges for the coffee can bottom to screw down to the grounded copper top of the moderator, and that works a charm.
This precise-same setup also works with the B10 tubes, only change you need is to set the HV correctly for whatever tube you're using. I really like the resulting signal quality coming out, nicely out of any noise (a few volts signal) and a good flat baseline. And the low power drain....this is getting really close to "as good as it gets" if the doggone tube will just act right. Drives plenty of feet of coax just fine, no artifacts (on this fairly slow signal, anyway). There is in general no need to impedance match the coax until you start having risetimes that make the coax a good part of a wavelength, till then, it just looks like a small capacitor across the output of the thing, which it seems to drive just fine.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.